Money, and expertise plain and simple.
10 years ago no one knew *** they were doing, so you just re-engineerred the same crap over and over. Today you try and buy from someone else if you can, and focus on building the game, not the tech.
When they started seriously into this project they were an edmonton developer with no mmo experience. So they set up shop in texas, and hired a bunch of people who'd made MMO's before. And the first thing they realized is starting a game engine from scratch is a pile of money.
A game engine is not just graphics and physics. Those are the parts we tend to emphasize the most when teaching, because the other parts are covered elsewhere though. How do you build the back end? BioWare had no expertise in that (nor did LucasArts, the EA buyout came later), can you build a 'the client is the worldbuilder' game engine from scratch? Sure, I can (no, seriously, I can), but it's a huge undertaking and it would have added years to the process. How quickly can you prototype things? Assign tasks? Etc. You can see from the hero engine website what the licence agreement BioWare has. The 'normal' agremeent is simutronics takes 30% but runs the servers. Or for a undisclosed fee they'll hand you all the source code for everything and you can run your own servers. Right now the Data centres seem to be the same basic areas as Simutronics, but I'm betting EA went with the "we own everything" licence, because that's the only sensible one for a probject this size. But while they were developing they had access to multiple clusters they could work on, that someone else managed. They could start building content while they were figuring out all of the networking, rather than having to build them in serial.
Back in 2006 when they were really strarting up this project, hero engine was Really good cutting edge stuff. Writing your own costs a lot of money and they had no expertise in any of the disrtibuted systems or networking architecture. Even the world building tools in hero engine are a step up from the aura engine (though I don't know about they lycium engine).
The hero engines biggest problem is with collision detection. That's a fairly common problem for all CPU physics solvers (see havok), or they run pitifully slow. The goal with TOR wasn't to build the DA2 engine for the game. The PC mmo market can't handle that. You need to be able to run TOR with a computer that will run WoW. That necessarily equates to some serious compromises in what you could have done. On the other hand, you don't have to worry about things like voice chat, (hero engine supports vivox), and so on. They do the back end programmer stuff, you do the front end designer stuff. As a canadian game programmer I'm not exactly thrilled with BioWare setting up shop in a foreign country, but these things happen in this business, you go wherever you get the best tax breaks (austin, montreal, toronto, vancouver etc.).